Search

Additional file 5: Supplementary Figure S5. Clustering benchmarks on 64 two CAMI II Mouse gut datasets. The number of identified clusters with precision ��� 0.90 and different recall levels were shown. All programs were run with their default parameters. MetaDecoder with minimum sequence length setting to 1 Kb (MetaDecoder1000) was also added for benchmarking. Assessments were evaluated using AMBER (version 2.0.2).

Additional file 8: Supplementary Figure S8. Clustering benchmarks on a cohort of T2D study. The number of identified clusters with contamination ��� 0.05 and different recall levels were shown. All programs were run with their default parameters. MetaDecoder with minimum sequence length setting to 1 Kb (MetaDecoder1000) was also added for benchmarking. Assessments were evaluated using CheckM (version 1.0.13).

Additional file 7: Supplementary Figure S7. Clustering benchmarks on 24 Human Microbiome Project datasets. The number of identified clusters with contamination ��� 0.10 and different recall levels were shown. All programs were run with their default parameters. MetaDecoder with minimum sequence length setting to 1 Kb (MetaDecoder1000) was also added for benchmarking. Assessments were evaluated using CheckM (version 1.0.13).

Additional file 3: Supplementary Figure S3. Clustering benchmarks on two CAMI I datasets. The number of identified bins on CAMI I (A) Medium and (B) High Complexity dataset with different score levels were shown. All programs were run in multi-sample mode with their default parameters except for MaxBin2 on the CAMI I High Complexity dataset. MaxBin2 was run on a single merged dataset duo to the excessive runtime with multiple samples. MetaDecoder with minimum sequence length setting to 1 Kb (MetaDecoder1000) was also added for benchmarking. Assessments were evaluated using AMBER (version 2.0.2).

Additional file 4: Supplementary Figure S4. Clustering benchmarks on 64 two CAMI II Mouse gut datasets. The number of identified clusters with precision ��� 0.95 and different recall levels were shown. All programs were run with their default parameters. MetaDecoder with minimum sequence length setting to 1 Kb (MetaDecoder1000) was also added for benchmarking. Assessments were evaluated using AMBER (version 2.0.2).

Additional file 2: Supplementary Figure S2. Clustering benchmarks on a simulated dataset. The number of identified bins with different score levels were shown. All programs were run with their default parameters in both single-sample (A) and multi-sample (B) modes. MetaDecoder with minimum sequence length setting to 1 Kb (MetaDecoder1000) was also added for benchmarking. Assessments were evaluated using AMBER (version 2.0.2).

Additional file 12: Supplementary Figure S12. Euclidean distance distributions between contigs abstracted from the same genome (intra-genome) and from two different genomes (inter-genome). The 3,023 bacterial genomes downloaded from NCBI were randomly fragmented into contigs (1 Kb - 100 Kb), then one million simulations were performed. Two histograms denote the intra- and interdistances, respectively. Two curves represent Gaussian densities parametrized by the mean and variance of the intra- and inter- distances, respectively.

Additional file 6: Supplementary Figure S6. Clustering benchmarks on five CAMI II Human Microbiome Project datasets. The number of identified bins on (A) Airways, (B) Gastrointestinal tract, (C) Oral cavity, (D) Skin and (E) Urogenital tract dataset with different score levels were shown. All programs were run in multi-sample mode with their default parameters. MetaDecoder with minimum sequence length setting to 1 Kb (MetaDecoder1000) was also added for benchmarking. DASTool was carried out on two combinations: 1) MetaDecoder and MetaBAT2, 2) CONCOCT and MetaBAT2. Assessments were evaluated using AMBER (version 2.0.2).

Additional file 9: Supplementary Figure S9. Clustering benchmarks on a cohort of T2D study. The number of identified clusters with contamination ��� 0.10 and different recall levels were shown. All programs were run with their default parameters. MetaDecoder with minimum sequence length setting to 1 Kb (MetaDecoder1000) was also added for benchmarking. Assessments were evaluated using CheckM (version 1.0.13).

With the emergence of multi-touch interactive display platforms, large size multi-touch platform is developed greatly. Cooperation and conflict is the problem for multi-user operation on huge screen. In this paper, the problem of multi-user's cooperation is studied. Based on the FTIR multi-touch interactive surface a solution scheme for multi-user cooperative touch interaction is proposed, in which a camera is placed above the interactive desktop for hand detection, tracking and ownership definition. So hand tracking of different users is key technique in the proposed scheme. Camshift method is used for hands of multi-users tracking in this paper, in which Opencv and C++ are applied to implement the function above.